Direct observations of the first generation of luminous objects will likely become feasible over the next decade. The advent of the Next Generation Space Telescope (NGST) will allow imaging of numerous galaxies and mini-quasars at redshifts z>5. We apply semi-analytic models of structure formation to estimate the rate of multiple imaging of these sources by intervening gravitational lenses. Popular CDM models for galaxy formation yield a lensing optical depth of about 1% for sources at redshift 10. The expected slope of the luminosity function of the early sources implies an additional magnification bias of about 5, bringing the fraction of lensed sources at z=10 to about 5%. We estimate the angular size distribution of high-redshift disk galaxies and find that most of them are more extended than the resolution limit of NGST, roughly 0.06 arcseconds. We also show that there is only a modest redshift evolution in the mean surface brightness of galaxies at z>2. The expected increase by 1-2 orders of magnitude in the number of resolved sources on the sky, due to observations with NGST, will dramatically improve upon the statistical significance of existing weak lensing measurements. We show that, despite this increase in the density of sources, confusion noise from z>2 galaxies is expected to be small for NGST observations.